This commit modifes the constant folder to not handle equality comparisons b/w
infinity and non-infinity operands. In such comparisons, special floating point
types - Float80 and Float16, may come into play and pattern matching againt them
complicates the constant folding logic more than we'd like.
Previously, `isScopeAffectingInstructionDead` determined that
an otherwise satisfactory `load` was not dead if it was a `load [take]`.
The immediate effect was that dead `load [take]`s were not deleted. The
downstream effect was that otherwise dead graphs of instructions would
not be deleted. This was especially a problem for OSLogOptimization
which deletes a great deal of code.
Here, the InstructionDeleter is taught to compensate for the deletion of
such `load [take]` by inserting `destroy_addr`s in their stead.
rdar://117011668
Before move-checking values, complete the lifetimes of all the values
derived from them via copy, borrow, and move.
Collect all such values and their derived transitive values and then
complete the lifetimes of each, visiting the instructions which produce
them in post-order.
Once OSSALifetimeCommpletion runs as part of SILGenCleanup, this code
can be deleted.
We already use a complexity limit for other functions in AliasAnalysis.
This is a workaround for quadratic complexity in ARCSequenceOpts.
Fixes a compile time problem
rdar://114352817
The metatype param may or may not be removed from the callee during
specialization. If it is, there is only one argument, if it is not,
there are two arguments. None of that is relevant to this test case.
It is just verifying that the strong_retain occurs before the apply.
`Dictionary.init(dictionaryLiteral:)` was annotated
`@_effects(readonly)` which means among other things that it doesn't
release any references. Being an init, however, it consumes its
arguments, and so does in fact release.
rdar://114699006
IRGen crashed in case an enum, which has a 24 bit payload (e.g. three `UInt8`), is used as a payload of another enum, e.g. `Optional`, in a statically initialized global variable.
rdar://112823823
The main change here is in IRGen which needs to be able to emit constant enum values.
Use `emitValueInjection` to create the enum constant.
Usually this method creates code in the current function.
But if all arguments to the enum are constant, the builder never has to emit an instruction.
Instead it can constant fold everything and just returns the final constant.
Also, create statically initialized let-globals as constant global (`constant` instead of `global`).
For very large control flow graphs the markControllingTerminatorsLive can stack overflow.
Fix this by doing the work iteratively instead of recursively.
rdar://106198943
The `isEscaping` function is called a lot from ARCSequenceOpt and ReleaseHoisting.
To avoid quadratic complexity for large functions, limit the amount of work what the EscapeUtils are allowed to to.
This keeps the complexity linear.
The arbitrary limit is good enough for almost all functions.
It lets the EscapeUtils do several hundred up/down walks which is much more than needed in most cases.
Fixes a compiler hang
https://github.com/apple/swift/issues/63846
rdar://105795976
- #58975 switched many tests from XFAIL on linux to linux-gnu, so seven
fail on the Android CI and two natively. They are now explicitly excluded.
- #39605 added several C++ Interop tests, 11 of which fail on the Android CI,
so disable them for now.
- #42478 removed the @noescape attribute for the non-Android
SIL/clang-function-types tests, so I remove it for Android too.
- My pull #40779 moved the Swift pointer tags to the second byte, so
SILOptimizer/concat_string_literals.64 will need to be updated for that,
disabled it for now.
- Compiler-rt moved the directory in which it places those libraries for
Android, llvm/llvm-project@a68ccba, so lit.cfg is updated for that.
Instead of doing one or two non-iterative BackwardReachability runs,
do a single run of IterativeBackwardReachability. During that, pause
after discovery/local dataflow and use VisitBarrierAccessScopes to
determine which end_access instructions in the discovered region are
barriers. Add those instructions as kills to the dataflow. Finally run
the global dataflow.
Enables SSADestroyHoisting to hoist destroys over loops.
Addresses a correctness issue where access scopes which were open at
barrier blocks were not promoted to barriers, resulting in destroy_addrs
getting hoisted into unrelated access scopes.
lit.py currently allows any substring of `target_triple` to be used as a
feature in REQUIRES/UNSUPPORTED/XFAIL. This results in various forms of
the OS spread across the tests and is also somewhat confusing since they
aren't actually listed in the available features.
Modify all OS-related features to use the `OS=` version that Swift adds
instead. We can later remove `config.target_triple` so that these don't
the non-OS versions don't work in the first place.
Verified that lexical lifetimes DO NOT result in a method call to a weak
reference resulting in a strong reference to the object. Consequently,
even with lexical lifetimes enabled, it is still possible, within a
single scope, for the first method call to an object weakly referenced
to occur but for the second such call not to because the object will
have been deallocated.
Verified that when a __consuming method calls a function which takes a
closure that captures self weakly, self is not deallocated until the
call returns. (Note that this is a behavioral change from what occurs
when lexical borrow scopes are disabled; in that case, self is
deallocated before the call to the function.)
The effect of passing -enable-copy-propagation is both to enable the
CopyPropagation pass to shorten object lifetimes and also to enable
lexical lifetimes to ensure that object lifetimes aren't shortened while
a variable is still in scope and used.
Add a new flag, -enable-lexical-borrow-scopes=true to override
-enable-copy-propagation's effect (setting it to ::ExperimentalLate) on
SILOptions::LexicalLifetimes that sets it to ::Early even in the face of
-enable-copy-propagation. The old flag -disable-lexical-lifetimes is
renamed to -enable-lexical-borrow-scopes=false but continues to set that
option to ::Off even when -enable-copy-propagation is passed.
Verify that lexical-lifetimes and copy-propagation interact correctly to
keep objects named by variables alive in a couple interesting cases:
unsafe pointers, synchronization point calls.
Previously, both swift-frontend and sil-opt put lexical lifetimes behind
a flag named -enable-experimental-lexical-lifetimes. That's redundant.
Here, the experimental portion of the name is dropped.
Previously, TempRValueElimination would peephole simple alloc_stacks,
even when they were lexical; here, they are left for Mem2Reg to properly
handle.
Previously, SemanticARCOpts would eliminate lexical begin_borrows,
incorrectly allowing the lifetime of the value borrowed by them to be
observably shortened. Here, those borrow scopes are not eliminated if
they are lexical.
Added an executable test that verifies that a local variable strongly
referencing a delegate object keeps that delegate alive through the call
to an object that weakly references the delegate and calls out to it.
We need to run utils/swift-darwin-postprocess.py on every executable test on Darwin platforms to work around a dyld issue. (And to ad-hoc sign executables on platforms that need it.)
Add a %target-codesign step to stress tests that are currently missing it.
